I tested the branches and if nobody object, I plant to tag the current HEADS as Dune 2.4.2 RC1 within 7 days. If no blockers appear, I'll release within 14 days, arround 2017-06-09.
I tested the HEADs of the current releases/2.4 branches with a recent CMake and Clang 4.0.0, recent Autotools and GCC 7, and GCC 4 together with CMake 2.8.9 on Debian 7. I found the following issues:
Recent versions of METIS and SuperLU break our old code. This is no surprise. If someone wants to fix it, they should provide according merge requests and test this with ancient GCC 4.4.
The backported writeVectorToMatlab contains range-based for loops, which breaks with older compilers like GCC 4.4. This might be fixable, but as it does not break existing code, I would stick to it.
@gruenich I prepared a bunch of merge requests to update the mailing list addresses in !259 (merged), dune-grid!182 (merged), dune-localfunctions!45 (merged) and dune-istl!101 (merged). I made a mistake when preparing the merge request for dune-geometry, and ended up committing onto the release branch directly in commit dune-geometry@eae90f19, please revert if that is unacceptable. There is no 2.4.2-rc1 tag for dune-grid-howto, so I'm assuming you're not planning to include it in the release and did not prepare an MR for it.
Thanks for providing the patches, its much appreciated!
Concerning PDELab, I would help releasing a 2.4.2, but I have no write access. The current HEAD of the releases/2.4 branch is pretty stable and I use it a lot.